关键词｜Key Words

保 ID 的人像生成、AI 换脸、AI 证件照、免训参考图生成、扩散模型、图像生成

Zero-shot Identity-Preserving Generation, AI Face Swapping, AI Portrait, Train-free Generation with Reference Image, Diffusion Model, Diffusion-based Image Generation

效果展示｜Results

数字形象(证件照)｜Digital Persona(AI Portrait)

通义照相馆在人像相似度和美观度上持续迭代，实现行业领先的高保真AI数字形象生成。

Tongyi Camera continuously iterates on portrait similarity and aesthetic appeal, achieving industry-leading high-fidelity AI digital persona generation results.

参考图｜Reference Image

1_副本.png

4_副本.png

v1(2407)

1.png

4.png

v2(2409)

2.png

55.png

v3(2410)

3.png

6.png

我们对业界难点——儿童和老年人数字形象专项提升，实现儿童和老年人数字形象的领先生成质感。

We have specifically enhanced the challenging area of digital personas for children and the elderly, achieving a leading level of quality in generating highly realistic digital representations for these groups.

参考图｜Reference Image

1727418202626-b61f3970-3ae3-4366-933a-c0b6c67bae61_副本.png

default_Portrait_transparent_nainai-0_1.0_0.95_0.9_0.85.png

v3(2410)

1727418202626-b61f3970-3ae3-4366-933a-c0b6c67bae61.png

default_Portrait_transparent_nainai-0.png

参考图｜Reference Image

1727418256559-31c37101-d9ec-4eae-8925-c54db83baf9d_副本.png

default_Portrait_transparent_yeye-4_1.0_0.95_0.9_0.85.png

v3(2410)

1727418256559-31c37101-d9ec-4eae-8925-c54db83baf9d.png

default_Portrait_transparent_yeye-0.png

AI换脸｜Face Swapping

1727664565950-30603087-2043-4217-9cb8-d72c1c254b69.png

1727664574323-69509dd9-c72f-439a-8dca-0892fea6d406.png

1727664575760-e3b343bb-b2ab-48ef-af53-4906229f8f17.png

1727664594203-8ed69c09-4aaa-4e94-834c-12eb86fe8b48.png

1727664616125-a0726037-ad56-4eca-8ee2-c4b59e74a86b.png

1727664713049-458e8195-4407-4b0d-815c-390007a3b3d8.png

1727664720550-2a8f2a96-20f2-44a7-8cb5-bf4721fad618.png

1727664730812-1a313fb0-bfe1-4243-9003-9fa80488dfc8.png

1727664734535-3c58a4f5-5cea-402e-ae33-7885a770c517.png

1727664738615-a0d67233-f41c-4d65-b775-51b821e1f99b.png

能力展示｜Demonstration

我们展示通义照相馆的换脸能力在极端困难场景下的应用，包括大姿态、大表情、大遮挡等业界挑战性课题。

We showcase Tongyi Camera's face-swapping capabilities when applied to extremely challenging scenarios, including large poses, expressive facial changes, and significant occlusions, addressing industry challenges.

每组的三张图分别为：参考图、底版图和生成结果 The three images in each set are: reference image, template image, and generated result.